Today’s hard drives contain internal control circuitry that handles low-level functions. These functions can include how the drive spins and moves to access specific information.
Different Types of Hard Drive Interfaces
Your computer requests this information over various interfaces such as:
- SATA
- PATA
- IDE
- FireWire
- SCSI (pronounced “scuzzy”)
These interfaces are relatively modern: a separate controller on the PC was once responsible for the hard drives’ low-level functionality. These controllers were sufficient for low-performance drives, but as data access speeds advanced into the 1980s and electronics become more compact, the control circuitry necessary to read physical hard drive locations moved onto the drive itself. Engineers needed a new control method.
To meet this need, Shugart Associates developed the Shugart Associates System Interface (SASI) to provide control between a computer and hard drive. SASI became standardized under ANSI as the Small Computer System Interface (SCSI), to avoid naming an official standard after a specific company.
SCSI became the adopted standard by a wide variety of computing companies such as Adaptec, Apple, Sun, Atari, and Amiga. For its part, IBM PCs never widely adopted SATA, preferring the Advanced Technology Attachment (ATA) standard – named after IBM’s AT computer and released in 1984. The ATA interface was a workable solution for home use, and some higher-end workstations took advantage of this technology through add-on hardware for increased performance characteristics.
What is SCSI?
Traditional SCSI interfaces contain a ribbon cable that attaches from the controller to a peripheral, along with a Single Connector Attachment (SCA) plug featuring 50, 68, or 80 pins. Hard drives may be the first devices that come to mind when considering SCSI, but this interface can work with devices like:
- Printers
- Scanners
- CD-ROM drives
In a way, SCSI acted as a sort of “universal bus” long before USB existed as we know it today. The first connected hard drive or another device can create a daisy chain connection for up to eight devices on a narrow bus, or 16 devices on a wide bus controller.
Unlike the USB standard that emerged in the 1990s, jumpers or switches had to physically set a unique ID for each device. The device on the end of this chain required a physical termination block instead of a cable leading to another drive. For more involved computing applications, eight or even 16 devices may seem restrictive. Fortunately, you could further subdivide the unique IDs into logical unit numbers (LUNs). Using advanced devices like disk arrays, users could address each ID individually.
SCSI Speeds
The original SCSI-1 standard was capable of five million data transfers per second, and it eventually evolved up to 160 with Fast-160 (DT). Data transferred in an 8-bit parallel bus, so five million transfers per second equates to 5MB/s (also expressed 5Mb/s). Later SCSI generations would advance to a rate of 160 million transfers per second, taking advantage of the wide bus SCSI architecture to double the data transfer rate to 320MB/s. In 2003, another standard capable of 640 MB/s entered the market. Unfortunately, it never saw widespread adoption, as more advanced data transfer methods began to emerge.
Disadvantages of SCSI: SATA and SAS Interfaces Take Over
Into the mid-2000s, SCSI remained a viable technology for high-end applications like RAID servers, which arranged a number of disks for redundancy and performance benefits. This technology eventually gave way to new technologies like SATA and SAS, as well as more advanced USB and Firewire connections. Interestingly though, if you break out these acronyms, SATA stands for Serial Advanced Technology Attachment, a reference that goes all the way back to 1984’s AT computer and connection standard. SAS, for its part, stands for Serial Attached SCSI, hiding its SCSI roots behind an acronym of acronyms.
Comparing SAS vs SCSI shows some unique advantages for SAS over traditional SCSI:
- Serial operation. SAS technology transmits data as a serial string of 1s and 0s, not over eight or 16 lines. Parallel connections suffer from disadvantages such as inter-symbol interference and noise, and serial operation significantly improved these issues.
- Speed. SAS-1, which entered the market in 2004, is capable of a 3Gbit/s data transfer rate, while 2017’s SAS-4 boasts a staggering 22.5Gbit/s data transfer rate. This increased speed works out to be over 500 times the capability of the original SCSI-1.
- Communication. Along with speed, SAS uses expanders, which allows for communication between a theoretical limit of 65,535 devices. Compare that to traditional SCSI’s limit of 15 devices per channel (the controller itself uses up one of the 16 IDs). As a bonus, SAS doesn’t require a terminator on the last drive.
SAS appears to be the high-end data transfer solution for the foreseeable future. Conveniently, even with these expanded capabilities, SAS interfaces use the same set of commands as traditional SCSI, easing the transition between the two.
Many engineers view SATA as a successor to consumer-level ATA – or parallel ATA, which distinguishes it from the previous generation. SATA provides a lower- priced option for those who don’t need server-level performance for everyday computing tasks.
From the obsolete “dumb” drives, the SASI format emerged and eventually gave way to SCSI. So many iterations later, we enjoy today’s double-digit gigabit transfer rates and massively parallel drive clusters. As interface technology continues to evolve, we can still see traces of the original data gathering intelligence from the 1970s and 1980s.
Special thanks to John Jensen for his help on this article.